simple-completion-language-server

Allow to use common word completion and snippets for Helix editor

demo.mp4

Based on comment

Install

From source

From GitHub:

$ cargo install --git https://github.com/estin/simple-completion-language-server.git

From local repository:

$ git clone https://github.com/estin/simple-completion-language-server.git
$ cd simple-completion-language-server
$ cargo install --path .

Nix

You can install simple-completion-language-server using the nix package manager from nixpkgs.

Note

At the moment the package is only available on the unstable channel of nixpkgs.

# Add it to a temporary shell environment
nix shell nixpkgs#simple-completion-language-server
# Add it to your current profile
nix profile install nixpkgs#simple-completion-language-server

Note

The above instructions assume you have enabled the experimental features nix-command flakes. If you are unsure about what this means or how to check read here.

If you are on NixOS or are using home-manager you can do one of the following:

# NixOS configuration.nix
environment.systemPackages = [pkgs.simple-completion-language-server];

# or home-manager config
home.packages = [pkgs.simple-completion-language-server];
# This will let `hx` know about the location of the binary without putting it in your $PATH
programs.helix.extraPackages = [pkgs.simple-completion-language-server];

Configure

For Helix on ~/.config/helix/languages.toml

# introduce new language server
[language-server.scls]
command = "simple-completion-language-server"

[language-server.scls.config]
max_completion_items = 100           # set max completion results len for each group: words, snippets, unicode-input
feature_words = true                 # enable completion by word
feature_snippets = true              # enable snippets
snippets_first = true                # completions will return before snippets by default
snippets_inline_by_word_tail = false # suggest snippets by WORD tail, for example text `xsq|` become `x^2|` when snippet `sq` has body `^2`
feature_unicode_input = false        # enable "unicode input"
feature_paths = false                # enable path completion
feature_citations = false            # enable citation completion (only on `citation` feature enabled)


# write logs to /tmp/completion.log
[language-server.scls.environment]
RUST_LOG = "info,simple-completion-language-server=info"
LOG_FILE = "/tmp/completion.log"

# append language server to existed languages
[[language]]
name = "rust"
language-servers = [ "scls", "rust-analyzer" ]

[[language]]
name = "git-commit"
language-servers = [ "scls" ]

# etc..

# introduce a new language to enable completion on any doc by forcing set language with :set-language stub
[[language]]
name = "stub"
scope = "text.stub"
file-types = []
shebangs = []
roots = []
auto-format = false
language-servers = [ "scls" ]

Snippets

Read snippets from dir ~/.config/helix/snippets or specify snippets path via SNIPPETS_PATH env.

Default lookup directory can be overriden via SCLS_CONFIG_SUBDIRECTORY as well (e.g. when SCLS_CONFIG_SUBDIRECTORY = vim, SCLS will perform it's lookups in ~/.config/vim instead)

Currently, it supports our own toml format and vscode json (a basic effort).

Filename used as snippet scope (language id), filename snippets.(toml|json) will not attach scope to snippets.

For example, snippets with the filename python.toml or python.json would have a python scope.

Snippets format

[[snippets]]
prefix = "ld"
scope = [ "python" ]  # language id https://code.visualstudio.com/docs/languages/identifiers#_known-language-identifiers
body = 'log.debug("$1")'
description = "log at debug level"

Use external snippets collections from git repos

Configure sources in ~/.config/helix/external-snippets.toml (or via env EXTERNAL_SNIPPETS_CONFIG)

1. Declare list of sources by key [[sources]]

2. Optionally, declare paths to load snippets on current source by [[source.paths]].

Highly recommended to declare concrete paths with scopes. To explicit configure required snippets and its scopes.

Snippets suggestion filtered out by document scope.

Scope it's language id and not file extension by language server protocol.

If [[source.paths]] isn't specified for source then all files with extensions .json and .toml would be tried to load. Scope at that case would be equal filename. To apply snippets the filename must be one of known language id.

3. Run commands to fetch and validate snippets

# Clone or update snippets source repos to `~/.config/helix/external-snippets/<repo path>`
simple-completion-language-server fetch-external-snippets

# Try to find and parse snippets
simple-completion-language-server validate-snippets

Config format for external snippets

# first external source to load snippets
[[sources]] # list of sources to load
name = "source1" # optional name shown on snippet description 
git = "https://example.com/source1.git" # git repo with snippets collections

[[sources.paths]] # explicit list of paths to load on current source
scope = ["scope1", "scope2"] # optional scopes (language id) for current snippets
path = "path-in-repo/snippets1.json" # where snippet file or dir located in repo

[[sources.paths]]
scope = ["scope3"]
path = "path-in-repo/snippets2.json"

# next external source to load snippets
[[sources]]
name = "source2"  
git = "https://example.com/source2.git"

[[sources.paths]]
scope = ["scope1"]
path = "path-in-repo-of-source2/snippets1.json"

Example

Load python snippets from file https://github.com/rafamadriz/friendly-snippets/blob/main/snippets/python/python.json

File ~/.config/helix/external-snippets.toml

[[sources]] 
name = "friendly-snippets"  
git = "https://github.com/rafamadriz/friendly-snippets.git" 

[[sources.paths]] 
scope = ["python"]  
path = "snippets/python/python.json"  

Clone or update snippets source repos to ~/.config/helix/external-snippets/<repo path>

$ simple-completion-language-server fetch-external-snippets

Validate snippets

$ simple-completion-language-server validate-snippets

Unicode input

Read unicode input config as each file from dir ~/.config/helix/unicode-input (or specify path via UNICODE_INPUT_PATH env).

Unicode input format (toml key-value), for example ~/.config/helix/unicode-input/base.toml

alpha = "α"
betta = "β"
gamma = "γ"
fire = "🔥"

Validate unicode input config

$ simple-completion-language-server validate-unicode-input

Citation completion

Citation keys completion from bibliography file declared in current document. When completion is triggered with a prefixed @ (which can be configured via citation_prefix_trigger settings), scls will try to extract the bibliography file path from the current document (regex can be configured via the citation_bibfile_extract_regexp setting) to parse and use it as a completion source.

To enable this feature, scls must be compiled with the --features citation flag. 

$ cargo install --features citation --git https://github.com/estin/simple-completion-language-server.git

And initialize scls with feature_citations = true.

[language-server.scls.config]
max_completion_items = 20
feature_citations = true

For more info, please check #78

Similar projects

• erasin/hx-lsp
• metafates/buffer-language-server
• rajasegar/helix-snippets-ls
• quantonganh/snippets-ls
• Stanislav-Lapata/snippets-ls
• ...(please add another useful links here)

Useful snippets collections

• rafamadriz/friendly-snippets
• ...(please add another useful links here)